A complete database of base pairs and base triplets was constructed by computer modeling based solely on geometric principles. This database comprises all experimentally known base pairs and base triplets and a much larger number of base pairs and base triplets that have so far not been observed in nature. Various computer programs have been developed that identify base combinations with high intrinsic stabilities and that classify base combinations according to certain structural criteria such as the ability to overlap in the positions of the glycosidic bonds. We are now in the process of systematically testing the hypothesis that it is possible to form novel base triplets by using structurally similar base triplets in known nucleic acid structures as scaffold. In order to prove the feasibility of this strategy it is necessary to model each novel base triplet in the context of the overall RNA structure, e.g. to check for potential steric conflicts and for the disruption of other stabilizing interactions upon substitution of the known base triplet. This task can best be accomplished by interactive molecular modeling on computers with advanced graphics capabilities. These modeling studies are initially limited to already known three-dimensional RNA structures such as tRNAs although it is planned to extend these studies to the design of novel types of RNA structure.